No its a big bios bug if c10 is enabled
any drive on
PCIEX4
PCIEX1_1
PCIEX1_2
wont work at all in windows and doesnt even show
it shows in the bios but doesnt enumerate in windows due to errors on the pcie bus and stalls the windows boot by around 30 seconds


Gigabyte provided me with a fixed bios for it
F9j dated 28/05/20 (Has a Bios ID of GA so is unique to my support request and also makes it a pain to get off the bios..)

F9k also dated 28/05/20 doesnt have the fix

I dont seem to be geting very far with the 4kn fix its like talking to children
but they will fix it in the end mark my words..

So, if windows on disk and C10 enabled, you can't boot, or = 30 sec stall first?
I can fix the BIOS ID for you if you want? Tell me what you need it to be, and dump BIOS region with FPT
Yes, sorry, I know what you mean about trying to explain things. Just tell them in as short as possible way >> Add this setting, or change this setting/leave hidden, to this value = 4Kn fix

My boot drive is on an m.2 the m.2's work fine
just any pcie4x or 1x slots dont

but yeah if your os was on say the 4x pcie slot it wouldnt boot if C10 was enabled in the bios

They have said they have sent a request off to ami by the looks of it though unsure as it also says factory

My last but 1 reply
I have already given you enough info to replicate this in Email-ID: 838645-13

but very well i will give you further info

I have 4 drives in this pc that are user configurable to either 512e or 4kn

1 x Corsair MP510 480GB
1 x Toshiba XG3 THNSN5512GPU7 512GB
2 x Sabrent Rocket 1TB (yes the ones you linked in reply to Email-ID: 838645-16)

My Sabrent Rocket 1TB's have data on them at the moment so these have to stay 512e for now

but my other two drives dont have any data on them (Toshiba XG3 + Corsair MP510)

I have just set them to 4kn and sent an nvme secure erase command to them to set 4kn and clear out the old 512e partitioning

then proceeded to power off the pc
power it back up and partition the drives from windows

See attached images for the following
See Image 1 + 2 for setting 4kn
See Image 3 confirming via powershell with fstuil (b: = Corsair MP510 e: = Toshiba XG3)
See Image 4 for secondary confirmation that Corsair Force MP510 480GB is set to 4kn
See Image 5 for secondary confirmation that Toshiba XG3 THNSN5512GPU7 512GB is set to 4kn
See Image 6 note the missing Force MP510 & THNSN5512GPU7 Toshiba when set to 4kn
See Image 7 when the drives are set back to 512e both drive show back up


With regards to have i tested this in other pc's, as it goes yes i have

My upstairs pc which is very old and has an Intel DQ77MK Q77 Chipset & a Xeon E3 1290 V2
I wanted to add nvme support to it a few years ago as it doesnt support nvme out of the box due to a lack of efi option rom for nvme
to cut a long story short i inserted the NvmeExpressDxe into the firmware to add support
and it has been booting off of a Toshiba XG3 THNSN5512GPU7 256GB for the past few years in 512e mode

I tested earlier and my hack of a bios for the DQ77MK supports booting to 4kn nvme drives
something gigabyte seem to be having trouble with

So yes this need to be fixed at the bios level

My last reply to them

Question: did you even look at what i provided
there is no data to corrupt as once you set it to 4kn
the bios will refuse to allow you to set that drive as bootable
and it just disapears from the bootorder list
and a windows os will refuse to install on it

as i have said already there needs to be boot support added to the bios for 4kn nvme and nvme via usb bridge chips

please just pass this to the bios dev's
to be fixed


Last reply from Gigabyte
Hello,

Our team has sent the question to the original AMI factory for confirmation, it will take some time. We will revert to you once we have an update.

--------------------

Oh here's the fixed bios btw if you want them
if you want to flash back you either have to use the backup bios or the flasher i have in the 15-05-20 zip

i will put the fixed 4k nvme bios there also if and when they ever provide me with one

15-05-20 = serial port s3 bug fixed + PCIEX4 c10 fixed
28-05-20 = serial port s3 bug fixed + PCIEX4 c10 fixed + PCIEX1_1 + PCIEX1_2 c10 fixed

https://mega.nz/folder/QzpXyKJT#UUWU_7nN9W5iV55-wjEqRg

@Shonk - Thanks - So, last BIOS above is not fully fixed (c10 fixed + PCIEX1_1 + PCIEX1_2 c10 fixed)?

Did you give them the info I gave you, about that single settings change that fixes this?? >> "Rapid Start Store Search Type" >> Needs to be set to following by default >> "PCIE/AHCI/NVME Controller" (Instead of Intel AHCI/RAID Controller)

Not yet as im waiting to see what they come back with

everything is fixed on the one i linked apart from 4kn boot

i dont use rst btw i use ahci

@Shonk - Ohh, I thought you said all OK in above notes about the last BIOS version.
RST being used in RAID mode or not does not matter, it's AHCI functions will be applied, and that setting needs set as I mentioned (we confirmed this is the 4Kn boot fix in the other thread I linked you to before)

Bravo！ That‘s so beautiful!

@June Da Pek - Thank you very much! It's very time consuming work. Next one, I will do black blue and maybe white/orange

That's pretty cool and sorry to revive your thread but I somehow ended up here after becoming once again confused and forgetting whether 'true' or 'yes' was correct for useplatformclock and realizing they were still interchangeable, because wow people use true a lot even though the documentation specifies yes/no.

However, I have to comment on the testing you used. CPU-Z's timers are good for loooong checks of stability, by actually running them, not just looking at the QPC timer. It was version 1809 that caused that timer issue to make it show 10mhz. It is unrelated to anything else, and it is not fixable in the bios or anywhere else. They changed something fundamental with how the QPC timer reports its frequency in 1809. I am dualbooted on 1709 and 1809 right now and can go back and forth from 10mhz to 3.xx with the same exact bcd edits.

You aren't generally looking for the lowest number- just the most stable combination of tscsyncpolicy, useplatformclock, useplatformtick, disabledynamictick, and the hpet option in BIOS. You need to run a game or program that actually is heavily affected by its use or misuse to see changes- though for me, having the clock on and hpet on makes everything hell. This will be variable by CPU, version of windows, and the program. TimerTest 1.4 (there are other ways, but I like how it does the timer pull) is the quick and dirty way to do it, + I see some correlation in the cpuz scores themselves, but not always. YMMV, but my conclusion was always, and has been since, 'disable hpet and delete all the values, don't even bother with no, they are for debugging like ms says they don't help'.

quick and dirty test i did when i was setting up my stuff for potential 3080 this week.


*differences, added that in too quickly, my mistake

Since you didn't seem to know about the 10mhz thing it might be useful to know that, imo, nothing changed, the timers still react the same, it just says 10mhz instead of what it used to show and some people think it feels worse overall, I'm not so sensitive as that, but I can read numbers at least. My tests were not very thorough, but the answer just reaffirmed my prior tests that were much more thorough, in that 'it doesn't matter' unless you are using a program that HATES hpet and platformclock on', but my windows always did, so its unbearable to even test

This was on an intel board, 8700k @ 4.7core and 4.7 uncore (i think lol i kinda forgot but those cpu-z scores make sense if so) with hyperthreading on, ram at default xmp with CR1 3200mhz, 2070 super. (not a gigabyte board tho I specifically found one that had hpet on it... but that might be a myth soon)

I like the design btw wish msi boards got anything more than CLICKING

@felicity - Yes, you can use Yes or TRUE, either one works.

I do not care about, or get bothered by anything HPET related, so my testing is limited to showing user enabled or disabled.
This is because users cannot see the setting when I disable, so may not "Feel/Think" anything changed since they cannot physically change it in BIOS themselves, so this way of testing is a quick way to show which method BIOS/OS(via BCD)/Both = best initial simple result.
Thank you for your details testing and info, this will help others! None of that makes any sense to me, since I'm not bothered by HPET so not familiar with any of that.

10Mhz thing, changes, based on my testing, 1809 OS bug only or not, changes are reflected if BIOS disabled HPET vs BIOS NOT Disabled HPET. See the bottom set of my test image
Now, the actual proper value you'd see if not for that bug may be different with other OS, but it's not relevant to why and how I tested and showed those results/info.

Thanks! MSI BIOS GUI can be redone too, if you wanted to put the time into redoing all the images and finding where the text color codes are stored etc.

Just got this mobo, can you please reupload modded bios? (HPET default). Seems like all links are down...

@dartraiden - Thanks for heads up! Yes, sorry, Tinyupload the main host has been down for more than a week now, hopefully they come back soon!!
Sad to also see the other mirrors down now too!
I've added two new mirror links for each, check first post

Hope you like the new theme!

Moar bugs (in addition to those described in the first post):

# it's possible to enable SecureBoot at the same time with Setup Mode (when there are no keys at all):
Select "Setup Mode" in SB options.You cannot enable SB directly if keys are missing. but you can restore previously saved settings (saved when SB was turned on). Reboot and... mobo cannot start :D It wil rollback to Backup BIOS after several unsuccessful start attempts.
fixed in F9l

# Mouse does not work in "Provision Factory Details" enabling confirmation window ("Press yes to proceed")
fixed in F9l

# Long settings descriptions are cropped and remain cropped even if you drag and drop them
partially fixed in F9l

# If PTT (Protection Trust Technology) is disabled and you will restore a backup where this setting is enabled, it will remain disabled after restart

@dartraiden - Thanks for your findings, those secure boot things may be fixed in later BIOS, I am not sure, you'd have to talk about all that with Gigabyte after testing whatever the newer BIOS are that are out now. Same for PTT
On the description thing, yes, I noticed that, and some I've edited adding my own wording/comments too (and tried to work around that)
But different BIOS resolutions or screen sizes altered how it was wrapped, so I gave up because I could not find universal solution that worked at all BIOS resolutions

Connecting to monitor in different ways (HDMI/DP etc), and Secure Boot Enabled/Disabled CSM Enabled/Disabled all affect BIOS resolutions (These, due to vBIOS or UEFI GOP being used)
You can check the BIOS resolution changed by the fans being "squished" looking on the EZ Mode main page, and or by saving screenshot of BIOS (F12) and then compare resolutions of each image while taken in different modes
I gave up due to could not resolve the wrapping universally, nor could I find a way to add back the old option we used to have to be able to directly choose the BIOS UI resolution.

btw
> Network modules not updated due to this breaks Network BIOS config options no matter how the mod is done. I only tested updating both (UNDI + Boot Agent) at once, maybe only one causes this

EFI Intel Gigabit UNDI 0.0.24 - ok
EFI Intel Gigabit UNDI 0.0.27 and newer - breaks